BioAlchemy: Distilling Biological Literature into Reasoning-Ready Reinforcement Learning Training Data

Despite the large corpus of biology training text, the impact of reasoning models on biological research generally lags behind math and coding. In this work, we show that biology questions from current large-scale reasoning datasets do not align well with modern research topic distributions in biolo...

Ful tanımlama

Kaydedildi:

Detaylı Bibliyografya
Asıl Yazarlar:	Hsu, Brian, Gökdemir, Ozan, Siebenschuh, Carlo, Parrello, Bruce, Getty, Neil, Brettin, Thomas S., Stevens, Rick L., Foster, Ian T., Chia, Nicholas, Ramanathan, Arvind
Materyal Türü:	Makale
Baskı/Yayın Bilgisi:	2026
Konular:	Artificial Intelligence I.2.6; I.2.7; J.3
Online Erişim:	https://arxiv.org/abs/2604.03506 http://arxiv.org/abs/2604.03506
Etiketler:	Etiketle Etiket eklenmemiş, İlk siz ekleyin!

Internet

https://arxiv.org/abs/2604.03506
http://arxiv.org/abs/2604.03506

BioAlchemy: Distilling Biological Literature into Reasoning-Ready Reinforcement Learning Training Data

Internet

Benzer Materyaller